The recent surge in conversational AI has opened up new avenues for the application of Behavioural Data Science, with Generative AI, particularly LLM models such as ChatGPT, representing a promising platform for analysing human behaviour. This chapter provides an overview of the role of Generative AI models in Behavioural Data Science, highlighting their strengths and limitations. The chapter begins by introducing the principles of Generative AI models and their potential applications in Behavioural Data Science. It discusses the advantages of using Generative AI models to study human behaviour, such as their ability to analyse large volumes of unstructured data and their capacity to learn from interactions with users. The chapter then describes the different ways in which Generative AI models can be used in Behavioural Data Science, such as analysing sentiment, predicting behaviour and generating insights from user interactions. It discusses the challenges associated with using Generative AI models, such as the need for accurate training data and the potential for bias in the model. The chapter also addresses the ethical concerns associated with the use of Generative AI models, such as privacy violations and the potential for unintended consequences. It discusses ways to address these concerns, such as implementing transparency and explainability in Generative AI models. The chapter concludes by discussing the future of Generative AI models in Behavioural Data Science, highlighting the potential for interdisciplinary collaboration with other fields such as psychology and sociology. It emphasises the need for continued development and refinement of Generative AI models and their associated methods for studying human behaviour. This chapter provides an overview of the role of Generative AI models in Behavioural Data Science, highlighting their strengths and limitations. It serves as a valuable resource for researchers and practitioners in the field who are interested in utilising Generative AI models to analyse and understand human behaviour.

Individuals have a surprisingly high capacity for making decisions quickly and still considering a multitude of information. This capability – often referred to as intuition – relies on automatic processes that can be described with neural networks. Particularly parallel constraint satisfaction (PCS) networks – a specific type of interactive activation networks – have been successful in capturing multiple aspects of choice behaviour. PCS models include restrictions to neural networks that capture specific features of cognition. This chapter will describe how PCS and other content models of decision-making can be evaluated and potentially improved by using artificial intelligence, specifically generic multi-layer (deep learning) neural network models. It will exemplify how choice behaviour can be modelled and predicted with PCS. The predictive performance of PCS will be contrasted with that of a generic neural network model. Possibilities and implications for the improvement of content models for choice behaviour using artificial intelligence are discussed.

This chapter examines the ethical considerations shaping Behavioural Data Science prior to the implementation of the European Union’s Artificial Intelligence Act. While recent regulatory developments mark a significant shift in legal obligations, many of the ethical challenges addressed here remain foundational and continue to inform practice today. Artificial intelligence and data science offer powerful tools for accelerating discovery across disciplines by enabling large-scale data analysis, uncovering behavioural patterns and revealing societal trends previously inaccessible to researchers. Yet the increasing reliance on data-driven methodologies introduces complex ethical risks – to individuals, communities and the broader social fabric – especially when behavioural insights are extracted from sensitive or repurposed data. This chapter traces how research communities and policy-makers, prior to the AI Act, developed and operationalised ethical data governance principles through various mechanisms. Among these, data trusts emerge as a particularly promising model. Situated between open access and rigid institutional control, data trusts provide a flexible framework for balancing the imperative to share behavioural data with the need to protect rights and foster public trust. The chapter explores the conceptual grounding and practical deployment of data trusts, arguing that, while formulated pre-AI Act, they remain highly relevant for ensuring legitimacy, accountability and transparency in behavioural data governance today.

This chapter provides an overview of Behavioural Data Science and categorises it into three strands: Human Behaviour, Algorithmic Behaviour and Systems Behaviour. The Human Behaviour strand seeks to understand and predict human behaviour using large datasets in a wide variety of applications. The Algorithmic Behaviour strand seeks to improve the performance of algorithms by studying how algorithms behave as well as how they process data about humans and systems, predicting future patterns. The Systems Behaviour strand studies how humans and algorithms collaborate in complex systems. The chapter also highlights the challenges and limitations of Behavioural Data Science, including ethical considerations and potential biases in data interpretation. It concludes with a discussion of the future of Behavioural Data Science and its potential for further advancements.

Cybersecurity remains one of the most technically complex domains of contemporary risk management, yet much of its vulnerability stems not from systems themselves but from the behavioural patterns of their users and adversaries. This chapter introduces a Behavioural Data Science perspective on cybersecurity – an approach that foregrounds the cognitive, social and affective processes underlying digital decision-making and threat response. Moving beyond the dominant paradigm of technological hardening, this chapter explores how insights from behavioural science, combined with computational modelling, can illuminate the conditions under which individuals make secure or insecure choices. It develops the concept of behavioural cyber-risk modelling, investigates the role of human–machine asymmetries and critiques static compliance-based regimes in favour of adaptive, context-aware systems. By synthesising empirical research and theoretical developments, the chapter proposes a new agenda for cybersecurity rooted in the nuanced realities of human behaviour, data-driven learning and interdisciplinary engagement.

Understanding the sources and consequences of luck has important behavioural and policy implications. Most prior research has treated luck as the residue of rationality, models or foresight. This chapter proposes a novel approach to help quantify the impact of luck and its interaction with human behaviours, particularly useful for behavioural data scientists. It illustrates the approach using three datasets with contexts that generate idiosyncratic patterns and behavioural implications, labelled swing luck, undeserved luck and network-bounded luck. The chapter concludes by discussing how this approach can be applied to other contexts and its scope conditions.

Behavioural Data Science is an interdisciplinary field at the intersection of behavioural science, data science, computer science, statistics, engineering and economics. This chapter offers a historical overview of the field’s evolution –from foundational developments in early computation and decision theory to the modern integration of machine learning, artificial intelligence and digital behavioural data. We explore major milestones, methodological innovations and successful applications across health, education, business and public policy. In addition to tracing this growth, we address the complex challenges the field faces, including ethical concerns, data limitations and interpretability of algorithmic models. We demonstrate how these formative tensions shaped the field’s development and continue to inform its future trajectory. The chapter concludes with a forward-looking perspective on how emerging technologies such as generative AI and quantum computing may further refine the goals, tools and ethical frameworks of Behavioural Data Science.

Machines are becoming more autonomous and intelligent, capable of making decisions and interacting with their environment. As they become more ubiquitous in our daily lives, it becomes increasingly important to understand and model their behaviour. This chapter will discuss the concept of machine behaviour and its importance in various fields, including artificial intelligence, robotics and human–computer interaction. The chapter starts by defining machine behaviour and providing an overview of its key characteristics. It will then explore the different approaches to modelling and understanding machine behaviour, including rule-based systems, statistical models and deep learning techniques. The chapter also covers the challenges and limitations of modelling machine behaviour, including the black box problem, interpretability and ethics. It also focuses on the applications of machine behaviour in various domains, such as autonomous vehicles, robotics and cybersecurity. The chapter will highlight how machine behaviour models can help in improving the performance, safety and security of these systems. The chapter discusses the future of machine behaviour and its potential impact on society. The chapter will explore the ethical implications of autonomous machines, including issues of responsibility, accountability and transparency. It will also discuss the need for interdisciplinary research in machine behaviour.

On the surface, behavioural data science and physics seem to be two disparate fields of research. However, a closer examination of problems solved by them reveals that they are uniquely related to one another. Exemplified by the theories of quantum mind, cognition and decision-making, this unique relationship serves as the topic of this chapter. Surveying the current academic journal papers and scholarly monographs, we present an alternative vision of the role of quantum mechanics in the modern studies of human perception, behaviour and decision-making. To that end, we mostly aim to answer the ‘how’ question, deliberately avoiding complex mathematical concepts but developing a technically simple computational code that the readers can modify to design their own quantum-inspired models. We also present several practical examples of the application of the computation code and outline several plausible scenarios, where quantum models based on the proposed do-it-yourself model kit can help understand the differences between the behaviour of individuals and social groups.

This chapter discusses the importance of beauty in our lives and how it has been under-valued until recently. Despite philosophers having written extensively on aesthetics, large scale quantitative evidence supporting the notion that beautiful places benefit humans has been limited to aspects of the environment that were traditionally measurable, such as the percentage of green land cover, tree density and different types of land cover via satellite imagery or population density via the census. However, behavioural data generated through our increasing interactions on the Internet has allowed us to quantify aspects of the visual environment that were previously difficult to measure. Using geotagged images uploaded to the Internet and advanced deep learning methods, we extract information about the aesthetic qualities of different locations. We use crowdsourced data from the online game Scenic-Or-Not, which rates geotagged photographs on the basis of how scenic they are, to investigate whether individuals achieve greater levels of happiness when encountering more scenic environments during their everyday life experience. We find that individuals are happier in more scenic locations, even in built-up areas, after taking other environmental measures such as green space into account. We show that inhabitants of more scenic environments report better health, even when taking core socioeconomic indicators of deprivation into account.

This chapter focuses on how AI influences arbitrators’ core tasks and decision-making – a development often described as ‘centaur arbitration’ – while recognising that, for now, the widespread use of AI by counsel is the main driver pushing this evolution forward. We address three central questions: (a) How do arbitrators currently use AI, and how might this develop in the future? (b) What challenges arise, and how do emerging guidelines seek to address them? (c) How does the use of AI by all arbitration participants affect the tribunal’s role and the balance between party autonomy, due process, and efficiency? By examining current practices alongside likely future developments, this chapter offers insights for practitioners and policymakers navigating the rapidly changing intersection of AI and arbitration. It shows how AI may redefine the tribunal’s responsibilities and reshape relationships within proceedings, highlighting the need for proactive regulation and thoughtful adaptation. Rather than advancing a normative conclusion, we aim to encourage reflection on how technological progress may influence our understanding of fairness, justice, and procedural integrity in arbitration.

Access to justice is a critical element of the rule of law, ensuring individuals can exercise their legal rights and resolve disputes through formal and alternative mechanisms. In the Netherlands, the judiciary plays a vital role in facilitating this access, but challenges remain, particularly for vulnerable groups. This paper introduces voorRecht-rechtspraak, an innovative online dispute resolution platform designed to address three key barriers to access: the presumption of citizen self-reliance, limited accessibility of legal aid, and the high costs of legal proceedings. Through a user-centred design and the integration of artificial intelligence (AI), voorRecht offers tools to support self-resolution of disputes while also providing structured human assistance for more complex cases. AI-driven features, such as simplified case-law summaries and semantic search functionality, improve the accessibility of legal information for non-experts, empowering citizens to engage with the law more effectively. While voorRecht is still in an iterative phase of development, early insights highlight its potential to reshape access to justice in the Dutch legal system.